Score contribution per author:
α: calibrated so average coauthorship-adjusted count equals average raw count
The purpose of this paper is to construct a leading index for the United States by deriving a set of weights based on Kalman filters. The weights have certain optimality properties and are related to the existing weighting methods of A. F. Burns and W. C. Mitchell (1946), S. H. Hymans (1973), and A. J. Auerbach (1982). The Kalman filter leading index is compared with the CIBCR leading composite index and an index suggested by Auerbach by subjecting all indexes to a number of tests. The results of the tests are mixed, but suggest that the use of Kalman filters as a way of constructing leading indexes is encouraging. Copyright 1990 by MIT Press.